Published in the June/July 2020 Issue – The analysis of data, who to distribute it to and how it will be used will be key to winning future conflicts.
On 18 February, the UK Ministry of Defence (MoD) announced moves to modernise the capabilities of the armed forces thereby allowing them to better operate throughout the ‘Information Age’.
Addressing delegates at the Royal United Services Institute’s (RUSI) inaugural Strategic Command conference in London, Minister for the Armed Forces, James Heappey described a series of challenges across both contemporary and future operating environments.
“It’s no longer enough to have a battle-winning edge in terms of fire power,” he warned. “There’s a responsibility to win the information battle. It’s no longer enough to have highly complex systems; you need all of the data that comes from that system in order to get a better understanding of what the enemy is doing and what the opportunities are to exploit and win the battle.”
Supported by key service leaders from the Royal Navy as well as industry, the minister also discussed how the MoD could “counter adversaries in the so-called ‘grey zone’ through special operations harnessing disruptive technologies such as artificial intelligence (AI) and big data in the cyber domain”.
Also speaking at the event was General Sir Patrick Sanders, Commander of the Strategic Command which comprises a rebranding of Forces Command. Sanders explained his intention to “strengthen the foundations of integration within the current force and experiment and develop the capabilities and structures required for the 2030s and beyond”.
“This will be achieved in three priority areas: cyber, special forces and multi-domain integration, all are transformative, all are essential,” he suggested.
As a result, the development and employment of AI and Machine Learning (ML) algorithms continues to be explored by Special Operations Forces (SOF) not only in the UK but around the world with commanders seeking to optimise the technology to support a range of mission sets including Intelligence, Surveillance, Target Acquisition and Reconnaissance (ISTAR) and Situation Awareness (SA).
One of the leading global entities investing deeply in such efforts is the US Special Operations Command (USSOCOM) which in October 2019 conducted its first ever Artificial Intelligence Symposium at MacDill Air Force Base to discuss emerging technologies, trends and capabilities.
The effort was the first tri-service meeting of the Command which was tasked with envisaging an ‘AI-enabled future’ for the USSOCOM.
The symposium followed a similar event conducted by the US Army Special Operations Command as well as the opening of USSOCOM’s Data Engineering Laboratory (DEL) at the SOFWERX facility in Tampa, Florida on 25 September.
At a ribbon-cutting, USSOCOM’s Commander General Richard Clarke discussed how ‘data’ will impact operations and investment in the future before outlining how the Command must move forward in attracting leading data scientists, architects, software developers and system integrators from around the world to “…improve the efficiency and capability of special operations forces”.
According to an official statement from USSOCOM, the DEL will also connect to the wider US Department of Defense (DoD) to focus on the “design and development of advanced data techniques including AI, machine learning, and robotic process automation”.
“SOF, along with the Department of Defense, is dedicated to advancing our data architecture and analytical tools. We believe this DEL is one of many that will emerge in the future ecosystem across the DoD, the broader US government, and foreign partners,” Clarke suggested at the event while USSOCOM’s Chief Data Officer, David Spirk highlighted how DEL ‘production’ will be aligned with USSOCOM AI goals. “This has been a huge, huge capability improvement for us and this is a realisation of what this future is for us,” he said..
Specifically, the DEL will support a series of USSOCOM AI/ML projects including preventative maintenance measures in support of the 160th Special Operations Aviation Regiment (SOAR). These aim to predict component failure to optimise flying hours and maintain operational readiness and efficiency.
Additionally, the DEL is also expected to support USSOCOM’s Hyper Enabled Operator (HEO) concept which was launched in 2019 to provide enhanced SA, lethality, connectivity, mobility and survivability to SOF operators working in austere environments.
However, arguably one of the most critical SOF-specific mission areas likely to immediately benefit from the application of AI and ML remains intelligence, surveillance, target acquisition and reconnaissance (ISTAR) with programmes including USSOCOM’s Joint Geospatial Analytic Support Services (JGASS) II programme which is seeking to enhance geospatial imagery analysis in support of special operations.
According to the publication of a pre-solicitation on 13 February 2020 (USSOCOM is expected to publish a request for proposals by the third quarter of 2020), the JGASS II aims to “support enterprise level Processing, Exploitation, and Dissemination (PED) of imagery related intelligence utilising a variety of advanced geospatial analysis techniques through the development and operation of an enterprise geospatial architecture that includes multiple GEOINT systems and data sets”.
“Analysts will be expected to produce regional and/or extremely detailed analytical products to support special operations ‘Find, Fix, Finish, Exploit, and Analyse’ [F3EA] targeting methodology using full motion video; imagery; and geospatial analysis from air, space, ground, and maritime Intelligence, Surveillance, and Reconnaissance assets.
“Analysts will also be expected to perform traditional and advanced PED on electro-optical, infrared, radar, and still frame imagery. Further, Analysts will be expected to work with Measurement and Signature Intelligence and datasets tailored to support Special Operations,” official documents highlighted before suggesting this be achieved through AI, ML and ‘other emerging technologies’.
One industry partner already working with SOF partners across the DoD, as well as partner forces and governments around the world, is Earth Intelligence (EI) specialist Maxar Technologies which is already starting to exploit AI and ML to support a limited number of ISTAR-related mission sets.
In 2021, Maxar Technologies launches its latest six-strong EI satellite constellation- WorldView Legion – which will significantly enhance the ability of SOF to observe image intelligence (IMINT) in ‘high demand areas of interest’ (AOIs) anywhere in the world up to 15 times in a single 24 hour period. Today, in-service EI satellites including Maxar Technologies’ own legacy constellations, retain capacity to revisit AOIs several times in the same period of time.
Speaking to Armada International, a Maxar Technologies official described how end users would benefit not only from the rapid increase in revisit times over AOIs but also 30cm resolution and 8-Band VNIR multispectral imaging.
“WorldView Legion will dramatically enhance the operational effectiveness of end users as they seek to streamline decision-making processes,” the spokesperson explained before confirming how the combination of commercial technology and innovation including AI/ML, computer vision and data science, would “transform the ever-increasing volume of data into more manageable, valuable, and consumable products and analysis for more timely, meaningful decision-making”.
“We’re evolving our analytic capabilities and GEOINT solutions to empower customers to go beyond simply describing and diagnosing situations to begin predicting incidents and prescribing intervention,” the spokesperson continued before explaining how WorldView Legion will be able to support a range of special operations including ‘military mapping for mission support’ and ‘maritime domain awareness’.
“Planning special operations in remote regions or poorly mapped environments presents unique challenges and risks. It is critical that operators and allies have the latest geospatial information, including maps and visualisation tools, to plan, rehearse and execute missions. Maxar’s three-dimensional elevation datasets, mosaics and other mapping products will help enhance situational awareness and minimise risks in mission planning,” the spokesperson continued.
Defence sources described to Armada how IMINT with frequent revisit times and high resolution would assist SOF commanders in the planning and preparation of direct action and special reconnaissance operations anywhere in the world, similar to the US Joint Special Operations Command’s Operation Kayla Mueller to capture of kill ISIS founder Abu Bakr Al Baghdadi on 27 October 2019.
“A detailed understanding of the ground as well as pattern of life of a target compound and its surrounding area is absolutely critical to the successful execution of any special operation, particularly hostage rescue operations,” one defence source highlighted.
Another SOF-specific special operations support provided by WorldView Legion’s increased revisit and resolution rates including Maritime Domain Awareness where SOF units can be tasked to conduct counter-terrorism, counter-piracy, counter-narcotics and (combat) search and rescue missions.
As the Maxar Technologies’ spokesperson continued to explain to AI, “nations with offshore territory need to focus their patrolling and security efforts for challenges including illegal fishing, maritime pollution, piracy, smuggling, human and drug trafficking to name a few.
“With WorldView Legion, coordination across our radar and optical satellite, and existing VDS [vessel detection system] capabilities Maxar will help maritime agencies address the myriad of challenges across the maritime domain. By helping focus resources through systematically detecting and identifying suspicious maritime activities, SOF will be able to enhance their ability to evaluate, prioritise, and respond faster than ever before.
Despite promoting AI/ML support in these types of special operations mission sets, the spokesperson also warned how the community was not fully trusting of AI/ML capabilities.
“It is operationally effective today but not yet fully trusted,” he highlighted. “EI must bring back the data and form conclusions, leveraging ML in a time frame with WorldView Legion fast enough to react to something seen with one sensor that could be rapidly updated with a second sensor.
“AI processing is getting faster and faster, largely because of advances in Cloud technology, better hardware and algorithms. The time taken to extract valuable and actionable information from IMINT is getting impressively short,” the spokesperson added.
As a result, the international SOF community continues to push companies like Maxar Technologies to ‘shorten timelines’ associated with data processing and exploitation.
“WorldView Legion will push data into the Cloud to run algorithms at scale against the content and deliver out an end product to end users with additional capabilities included ‘Automated Change Detection’- another application which could act as a trigger for special operations.”
Elsewhere, Booz Allen Hamilton is also pressing ahead with the support of special operations with the continued development of its Modzy AI solution which has applications for special operations with rapid access, evaluation, deployment, embedding and management of AI models at scale.
Launched in November, Modzy provides customers with a platform and model marketplace for the upload, management and re-use of AI models to reduce risk in operations. Applications could include the exploitation of Open Source Intelligence (OSINT) to identify high value targets and military equipment in addition to ‘Overhead Building Detection’ and facial recognition- two more SOF-specific mission areas which Dr Josh Sullivan, senior vice president at Booz Allen Hamilton highlighted to Armada.
Sullivan described Modzy as an “open architecture software solution available to customers on-premise, in the cloud, or via custom deployments” providing API access, built-in governance, adversarial defense, and explainability, the latter of which is described as one of the “toughest challenges related to scaling trustworthy AI”.
“The Modzy platform can be used to govern and manage customers’ own models for special operations applications, with customers interested in its ability to solve the last-mile challenges of AI at scale. Some of the use cases we’ve heard from defence clients include unique models that provide computer vision capability, including an aerial building segmentation model that can detect, mask, label and return information about buildings found in satellite imagery in order to assess building damage more quickly and with better accuracy more to aid in disaster relief efforts.
“Or, a military equipment classification model that can analyse JPEG images and classify the images into one of 87 military equipment classes, making it easier for analysts to process large set of images into different military equipment categories,” Sullivan concluded.
Although the potential for AI/ML support of special operations remains strong in potential, much work must be undertaken by the international SOF community in terms of not only better understanding the technology but also providing the means for operators to truly trust it moving forward.
by Andrew White